1. Using ROOT classes from Python ROOT Workshop 14-16 October 2002 P. Mato / CERN

2. 8/3/02Using ROOT classes in Python2 Contents  Motivation  Examples  Why Python?  Software requirements, Design, Status  Integration using Python as a Software Bus  Summary

3. 8/3/02Using ROOT classes in Python3 Motivation  Be able to use any ROOT class from Python in a generic way. –Without the need of wrapping each class –Using the ROOT object dictionary information  Facilitate access of ROOT files and other facilities from non-ROOT applications  Proof-of-concept that Python can be viewed as Software Bus –In analogy to a “hardware bus” where you can plug a variety of modules and interface adaptors to other buses.

4. 8/3/02Using ROOT classes in Python4 C:\> python... >>> from rootmodule import * >>> f1 = TF1('func1','sin(x)/x',0,10) >>> f1.Eval(3) 0.047040002686622402 >>> f1.Derivative(3) -0.34567505667199266 >>> f1.Integral(0,3) 1.8486525279994681 >>> f1.Draw() : created default TCanvas with name c1 Example  No much difference between CINT and Python !

5. 8/3/02Using ROOT classes in Python5 Why Python?  Interpreted –It is quite fast (byte code idea from Java)  Dynamically typed –No need to declare any variable  Simple syntax –Emphasis by the authors to minimize typing  Variety of available shells  Powerful built-in types and modules  Ideal for Scripting and Prototyping  Very popular nowadays (many users, many books(>20), many available modules (~1500), etc.)

6. 8/3/02Using ROOT classes in Python6 Python Language  Features: –Variables & Arithmetic expressions –String manipulations –Conditionals (if/else statements) –Loops (for/while) –Functions –Lists –Dictionaries –Classes (Objects) –Exceptions –Modules

7. 8/3/02Using ROOT classes in Python7 Another example import sys, string infile = open( 'aptuple.txt', 'r' ) lines = infile.readlines() title = lines[0] labels = string.split( lines[1] ) from rootmodule import TFile,TNtuple outfile = TFile('aptuple.root','RECREATE',‘NTuple file') ntuple = TNtuple('ntuple', title, string.join(labels,':')) for line in lines[2:]: words = string.split( line ) row = map( float, words ) apply( ntuple.Fill, row ) outfile.Write() CERN Personnel Category Division Flag Age Service … 202 9 15 58 28 0 10 13 4 40 11975 530 5 15 63 33 0 9 13 3 40 10228 316 9 15 56 31 2 9 13 7 40 10730 … makerootntuple.py For each line in the file, split the line into a list of words. Convert the list of words into a list of floating-point numbers. And finally, call the Fill() method of the ntuple with it.

8. 8/3/02Using ROOT classes in Python8 Software requirements  Generic Interface –Make use of the generic python functions (__call__, __getattr__, …) to give illusion that all types are known to Python –Use ROOT dictionary information at run-time (perhaps sacrificing some performance). No need to generate any code/library.  Emulating ROOT look and semantics while being Python natural –Same ROOT class names, method names, global variables, etc. –Python object instantiation, namespaces, garbage collection, etc.  The “rootmodule” extension module should be self- contained –Nothing else should be required except rootmodule.so[.dll]

9. 8/3/02Using ROOT classes in Python9 Design TClassTClassWrap TMethodWrap TObjectWrap TCArrayWrap TObject TMethod RootModule CINT Boost.Python Other ROOT Libraries TROOTWrap Python interpreter

10. 8/3/02Using ROOT classes in Python10 Boost.Python  Boost.Python library helps exporting C++ classes to Python.  It is designed to be minimally intrusive on your C++ design.  You need to reflect the C++ classes and functions into Python python::class_builder tobject(this_module, "TObjectW"); tobject.def(&TObjectWrap::getValue, "__getattr__"); tobject.def(&TObjectWrap::getName, "GetName"); tobject.def(&TObjectWrap::className, "ClassName"); tobject.def(&TObjectWrap::members, "members"); tobject.def(&TObjectWrap::repr, "__repr__"); tobject.def(&TObjectWrap::del, "__del__"); Python names C++ class/method

11. 8/3/02Using ROOT classes in Python11 Boost.Python (cont’d) >>> f1 = TF1('func1','sin(x)/x',0,10) >>> dir(f1) ['ClassName', 'GetName', '__del__', '__doc__', '__getattr__', '__module__', '__repr__', 'members'] >>> type(TF1) >>> type(f1) >>> type(f1.Draw)  From the point of view of Python all classes are of type TClassW, all objects are of type TObjectW, all methods are of type TMethodW

12. 8/3/02Using ROOT classes in Python12 Mapping ROOT from Python  Types –Trivial mapping of basic types (float, string, …) –ROOT Classes into instances of TClassW python class –Not yet mapping of TCollection into python built-in types (list?, dictionary?).  Global variables –Created instances with names gROOT, gSystem, etc.  C-arrays –ROOT uses C arrays (e.g. float[], float*) and Python can not handle them. Created special type to handle it.

13. 8/3/02Using ROOT classes in Python13 Known Problems  The overall performance on calling methods is not optimal. –This is mainly due to the overheads in argument conversions. (Python types->C++ types->text string-> CINT interpretation->method) –Using the CINT dictionary directly should improve it.  Calling python from CINT is not working if in different thread. –E.g. GUI call-backs on Windows  Not all the basic types are probably supported yet. Specially of the "C" like types like arrays or pointers.

14. 8/3/02Using ROOT classes in Python14 Known problems (cont’d)  Naïve memory management (object ownership). ROOT objects created by Python are deleted by Python.  The ROOT classes are not yet 100% equivalent to native Python classes, therefore some built-in functionality in shells (e.g. command completion) is not working.

15. 8/3/02Using ROOT classes in Python15 It is available for testing  The code can be downloaded from http://cern.ch/Gaudi/RootPython for tests http://cern.ch/Gaudi/RootPython  It currently uses –Boost 1.27, Python 2.2, ROOT 3.03  It is built using CMT for the time being  Some python scripts examples exists at …/tutorials

16. 8/3/02Using ROOT classes in Python16 Python as a Software-Bus  Python could also be seen as a framework where you can plug easily “extension modules” in binary form implemented using other languages. –Very easy and natural to interface to C++ classes (C++ API) –Python should only be the “glue” between modules developed in C++ or other languages –The interface (API) for Python extension modules is quite simple and at the same time very flexible (generic functions)

17. 8/3/02Using ROOT classes in Python17 Module Communication  Communication between two extension modules is always possible using Python primitive types (a) –Example: I would like to display an histogram of Gaudi Gaudi (C++)Lizard (C++) Python IH1D IH1D* Plotter Python Primitives  It is more efficient to communicate using a reference to an interface (AIDA?) (b) –The objects in Python are only references (a) (b) Both ways are possible!!

18. 8/3/02Using ROOT classes in Python18 GUI What is Feasible Python math shell gaudipython DatabaseEDG API GUI Very rich set of Python standard modules Several GUI toolkits XML Very rich set specialized generic modules Gaudi Framework rootpython Root Classes PVSS JPE Java Classes LHC modules Gateways to other frameworks

19. 8/3/02Using ROOT classes in Python19 Example Filling an Excel spreadsheet from a ROOT ntuple # Get the ntuple from the ROOT file import rootmodule hfile = rootmodule.TFile('hsimple.root') ntuple = rootmodule.gROOT.FindObject('ntuple') entries = ntuple.GetEntries() nvar = ntuple.GetNvar() tuple = ntuple.GetArgs() # Initialize Excel import win32com.client excel = win32com.client.Dispatch('Excel.Application') wbook = excel.Workbooks.Add() wsheet = wbook.WorkSheets.Add() wsheet.Name = ntuple.GetTitle() # Fill Excel sheet for i in xrange(500) : ntuple.GetEntry(i) for j in range(nvar) : wsheet.Cells(i+1,j+1).value = tuple[j] # Make Excel sheet visible excel.Visible = 1

20. 8/3/02Using ROOT classes in Python20 Summary  Exercise to proof that is possible to interface ROOT with Python with very little effort. –Demonstrate the power of having an object dictionary  The module is functional and it is already in use to provide access to ROOT files from other applications (e.g. HippoDraw) –Of course, more work is needed to make it complete and fix the current performance problems  Step in the direction of high level integration using a “software-bus” –One more Python “gateway” to other frameworks or languages like DCOM, Java, Gaudi, etc.